An Automatic Calibration and Feedback Method and System

Abstract

The present application provides an automatic calibration and feedback method and system; the automatic calibration and feedback method includes: collecting coordinate information of a calibration block in a preset pose; then obtaining mechanical information based on each coordinate information to calibrate a target product according to the mechanical information; further collecting an image of the calibrated target product, and obtaining a calibration result based on the target product image to feedback compensation information according to the calibration result. The calibration steps of the present application are fewer, the calibration process is simple and the calibration efficiency is high. At the same time, it can also reduce the calibration difficulty, improve the calibration accuracy, and form a closed-loop control with adaptive feedback, realizing real-time tracking and feedback of data and real-time adjustment of abnormal data; and the whole process is automatically completed, reducing manual intervention, reducing operation errors, and greatly improving production efficiency and quality; in addition, this method is applicable to various different target products and production environments, and has high flexibility and versatility.

Description

Technical Field

[0001] This application relates to the field of calibration technology, and particularly to an automatic calibration and feedback method and system. Background Art

[0002] In the field of automation, product positioning is divided into mechanical positioning and vision positioning. To improve the automation rate, reduce the debugging difficulty of the mechanism, and implement a general mechanism solution, the vision-based positioning method is widely used in various fields of automation. Its greatest significance lies in compensating for the defects that mechanical positioning cannot compensate for the differences in product incoming materials, greatly improving the data stability of products and ensuring the excellent rate of products.

[0003] However, for the vision-based positioning method, to ensure the positioning accuracy, it is necessary to calibrate the coordinate relationship between the camera and the actuator. Currently, the relatively mature method is the manual calibration method, which is difficult, time-consuming and laborious, while the automatic calibration technology is relatively weak and the accuracy is difficult to guarantee. Summary of the Invention

[0004] This application provides an automatic calibration and feedback method and system to solve the technical problems of the large difficulty, time-consuming and laborious caused by the existing manual calibration method, as well as the relatively weak existing automatic calibration technology and the difficult-to-guarantee accuracy.

[0005] Specifically, this application provides an automatic calibration and feedback method, including the following steps:

[0006] S100: Collect the coordinate information of the calibration block in a preset pose.

[0007] S200: Obtain mechanical information based on each coordinate information to calibrate the target product according to the mechanical information.

[0008] S300: Collect the image of the calibrated target product, obtain the calibration result according to the target product image, and feedback compensation information according to the calibration result.

[0009] In the above technical solution, an automatic calibration and adaptive feedback method is provided. The calibration steps are few, the calibration process is simple and the calibration efficiency is high. At the same time, it can also reduce the calibration difficulty, improve the calibration accuracy, and form a closed-loop control with the adaptive feedback, realizing real-time tracking and feedback of data, and adjusting abnormal data in real time; and the whole process is automatically completed, reducing manual intervention and operation errors, and greatly improving the production efficiency and quality to a certain extent; in addition, this method is applicable to various different target products and production environments, and has high flexibility and versatility.

[0010] Further, before performing step S100, it includes:

[0011] Under the first vision module, the calibration block is moved by an actuator, and the preset secondary pose is changed, so as to collect in real time the pose images corresponding to the preset secondary pose through the first vision module.

[0012] In the above technical solution, by changing the preset secondary pose and collecting the pose images in real time, the diversity and quantity of samples can be increased, providing more data for calibration; by sampling the pose multiple times, the error factors affecting pose measurement can be reduced, thereby improving the accuracy of calibration; by obtaining the pose images corresponding to the preset secondary pose, the coverage of coordinate information in different pose situations can be increased, thereby improving the reliability and stability of calibration; this method can adaptively change the preset secondary pose according to the actual situation to meet the calibration requirements to the greatest extent and adapt to different calibration working conditions.

[0013] Further, the step S100 includes:

[0014] Obtain the pose information in the first dimension, second dimension, and third dimension according to each pose image, so as to obtain the coordinate information corresponding to each pose image according to the pose information.

[0015] In the above technical solution, by obtaining the pose information in the first dimension, second dimension, and third dimension, the position and pose information of the target in the three-dimensional space can be completely described; this step uses machine vision technology to extract the pose information from the pose images, making the subsequent processing more convenient and efficient; by accurately extracting the pose and calculating the coordinate information, higher precision can be obtained, improving the accuracy and reliability of calibration; by collecting the pose information in different dimensions multiple times, the influence caused by errors can be reduced, improving the calibration accuracy.

[0016] Further, the step S200 includes:

[0017] Input each pose image and the corresponding coordinate information into a preset neural network model to obtain mechanical information; and position and fit the target product according to the mechanical information through the actuator.

[0018] In the above technical solution, by adopting a neural network model, the input pose image and coordinate information can be effectively processed and analyzed, so as to obtain accurate mechanical information, improving the efficiency and accuracy of calibration; the preset neural network model has automatic learning and adaptability, and can continuously improve the processing ability for various different coordinate information through training and optimization to adapt to different calibration scenarios; due to the fast calculation speed of the neural network model, real-time acquisition and feedback of mechanical information can be achieved, ensuring the application of the positioning and fitting of the target product in a fast production environment; adopting a neural network model can adapt to different target products. Whether the target product has shape changes, size changes or attitude changes, it can accurately identify and locate, improving the adaptability of the system; by automatically acquiring mechanical information and performing positioning and fitting by the actuator, the need for manual participation is reduced, and the complexity of the operation and the risk of human error are lowered.

[0019] Further, the acquisition of the target product image in step S300 includes:

[0020] Acquire the target product after positioning and fitting through the second vision module to obtain the target product image.

[0021] In the above technical solution, by acquiring the target product image after positioning and fitting, the accuracy of the positioning and fitting can be verified to ensure that the target product is in the correct position; the acquired target product image can be used for subsequent inspection and analysis. By comparing the differences with the ideal model, the quality of the positioning and fitting can be evaluated to further optimize the production process; and the image acquisition of the target product can be used for recording and archiving, forming a complete record of the production process, and serving as an important data basis for quality control and quality traceability.

[0022] Further, the feedback compensation information in step S300 includes:

[0023] Compare the target product image with a preset image to obtain a deviation value; if the deviation value is greater than a preset threshold, it is determined that the current calibration of the target product is unqualified, and the target product image is fed back to the preset neural network model for data compensation according to the target product image; otherwise, it is determined that the current calibration of the target product is qualified, and the current process ends.

[0024] In the above technical solution, by comparing with a preset image and calculating the deviation value, the system can automatically judge the calibration quality of the target product without manual intervention, improving the automation degree of the process; performing real-time deviation calculation and determination on the target product image can timely detect calibration problems, so as to take compensation measures in time and improve production efficiency; by judging whether the calibration is qualified according to the deviation value, the quality control of the production process can be carried out, timely detecting and handling calibration anomalies, and improving the product quality level; by feeding back the target product image to the preset neural network model for data compensation, the calibration process can be continuously optimized, improving the accuracy and stability of calibration.

[0025] Based on the same concept, the present application also provides an automatic calibration and feedback system, which includes:

[0026] An acquisition module: used to acquire the coordinate information of the calibration block in preset poses.

[0027] A calibration module: used to obtain mechanical information according to the acquisition result of the acquisition module, so as to calibrate the target product according to the mechanical information.

[0028] A feedback module: used to acquire the image of the calibrated target product, and obtain the calibration result according to the target product image, so as to feedback compensation information according to the calibration result.

[0029] In the above technical solution, an automatic calibration and feedback system is provided. While having fewer calibration steps and a simple and efficient calibration process, it can also improve the calibration accuracy and reduce the calibration difficulty; the system forms a closed-loop control with adaptive feedback, enabling real-time tracking, feedback of data, and real-time adjustment of abnormal data; the entire calibration process is completely automated, reducing manual intervention and operation errors, thereby greatly improving production efficiency and product quality to a certain extent; in addition, this method is applicable to various target products and production environments, with high flexibility and versatility.

[0030] Further, the acquisition module includes:

[0031] A first acquisition unit: used to acquire the preset pose images of the calibration block.

[0032] A second acquisition unit: used to obtain the pose information in the first dimension, second dimension, and third dimension according to each pose image, so as to obtain the coordinate information corresponding to each pose image according to the pose information.

[0033] In the above technical solution, by collecting the pose images corresponding to the coordinate information, the information of the calibration block in different poses can be accurately obtained, improving the accuracy of collection; the second collection unit can obtain the pose information in the first dimension, the second dimension and the third dimension, which can comprehensively reflect the position and pose information of the calibration block; through the automatic processing and analysis of the pose images, the automatic acquisition and calculation of the coordinate information can be realized, reducing the need for manual operation and improving the degree of automation.

[0034] Further, the calibration module includes:

[0035] The first acquisition unit: used to input each pose image and the corresponding coordinate information into a preset neural network model to obtain mechanical information.

[0036] The calibration unit: used to perform positioning and fitting on the target product according to the mechanical information.

[0037] In the above technical solution, through the preset neural network model and mechanical information, the calibration module can accurately calculate the position and pose information of the target product, and high-precision positioning and fitting can be achieved; the preset neural network model has self-adaptability and learning ability, and can be automatically adjusted and optimized according to different calibration scenarios and data inputs, so as to adapt to various different target products and calibration requirements; the calibration module uses the neural network model to automatically calculate and process the mechanical information, realizing the automatic operation of the calibration process, reducing the need for manual intervention, and improving the work efficiency; through the rapid calculation and adjustment of the calibration module, the time and cost of the calibration process can be reduced, and the efficiency of the production line can be improved.

[0038] Further, the feedback module includes:

[0039] The second acquisition unit: used to collect the target product after positioning and fitting to obtain the target product image.

[0040] The comparison unit: used to compare the target product image with a preset image to obtain a deviation value.

[0041] The feedback unit: used to, when the deviation value obtained by the comparison unit is greater than a preset threshold, feedback the target product image to the preset neural network model to perform data compensation according to the target product image.

[0042] In the above technical solution, the feedback module can quickly respond to the deviation of the target product and improve the accuracy of positioning and fitting through data compensation operations; since the feedback module can collect the target product image in real time and perform comparison and judgment, the feedback processing has high real-time performance and fast response ability; the preset neural network model has self-adaptability and learning ability and can automatically adjust and optimize according to the feedback data; the feedback module can quickly identify the deviation situation and perform data compensation processing, improving the efficiency and accuracy of the calibration system.

[0043] Compared with the prior art, the beneficial effects of the present application are as follows:

[0044] The present application first collects the coordinate information of the calibration block in a preset pose; then obtains the mechanical information according to each coordinate information to calibrate the target product according to the mechanical information; further collects the image of the calibrated target product and obtains the calibration result according to the target product image to feedback the compensation information according to the calibration result. The method proposed by the present application has fewer calibration steps, a simple calibration process and high calibration efficiency. At the same time, it can also reduce the calibration difficulty, improve the calibration accuracy, and form a closed-loop control with adaptive feedback, realizing real-time tracking and feedback of data and real-time adjustment of abnormal data; and the whole process is automatically completed, reducing manual intervention and operation errors, and greatly improving the production efficiency and quality to a certain extent; in addition, this method is applicable to various different target products and production environments, and has high flexibility and versatility. BRIEF DESCRIPTION OF THE DRAWINGS

[0045] Figure 1 is a flowchart of the automatic calibration and feedback method described in the present application.

[0046] Figure 2 is Figure 1 a system framework diagram of the automatic calibration and feedback method. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0047] The present application provides an automatic calibration and feedback method and system to solve the technical problems of high difficulty, time-consuming and laborious caused by the manual calibration method in the prior art, and the weak existing automatic calibration technology and difficult to guarantee the accuracy.

[0048] The following further describes in detail an automatic calibration and feedback method and system of the present application with reference to specific embodiments and the accompanying drawings.

[0049] Embodiment 1:

[0050] Please refer to Figure 1 , the present application provides an automatic calibration and feedback method, including the following steps:

[0051] S100: Collect the coordinate information of the calibration block in a preset pose.

[0052] Further, before performing step S100, it includes:

[0053] Under the first vision module, move the calibration block through the actuator and change the preset sub-poses to collect the pose images corresponding to the preset sub-poses in real time through the first vision module.

[0054] In this embodiment, the first vision module is a CCD vision system. The CCD vision system includes a camera, a lens, a light source, a light source controller, etc., and adopts a monocular vision method to collect image information as the pose of the actuator changes.

[0055] Among them, the CCD vision system has characteristics such as high precision, high resolution, high frame rate, and low noise. Its ability to collect, image, and transmit images is very good, and it can meet the requirements of high speed, high precision, and high efficiency in industrial automation.

[0056] The preset sub-poses are preferably 5 times. Those skilled in the art can also set according to actual calibration requirements, not limited to only 5 times.

[0057] The actuator grabs the calibration block and moves it under the CCD vision system, and changes the pose 5 times through the actuator, including rotation and movement of the x and y axes for operation. The CCD vision system collects the image information (i.e., the pose images) in different poses.

[0058] In the above technical solution, by changing the preset sub-poses and collecting the pose images in real time, the diversity and quantity of samples can be increased, providing more data for calibration; by sampling the pose multiple times, the error factors affecting pose measurement can be reduced, thereby improving the accuracy of calibration; by obtaining the pose images corresponding to the preset sub-poses, the coverage of coordinate information in different pose situations can be increased, thereby improving the reliability and stability of calibration; this method can adaptively change the preset sub-poses according to the actual situation to meet the calibration requirements to the greatest extent and adapt to different calibration working conditions.

[0059] Further, the step S100 includes:

[0060] Obtain the pose information in the first dimension, the second dimension, and the third dimension according to each pose image, so as to obtain the coordinate information corresponding to each pose image according to the pose information.

[0061] In this embodiment, the five pose images respectively include different pose information in the three dimensions of X, Y, and R to include all coordinate information.

[0062] The obtained coordinate information is, for example, P1(0, 0, 0), P2(-3, 3, 3), P3(3, 3, 3), P4(3, -3, -3), P5(-3, -3, -3); where P1 is the origin point.

[0063] In the above technical solution, by obtaining the pose information in the first dimension, second dimension, and third dimension, the position and pose information of the target in the three-dimensional space can be completely described; this step uses machine vision technology to extract the pose information from the pose image, making subsequent processing more convenient and efficient; by accurately extracting the pose and calculating the coordinate information, higher accuracy can be obtained, improving the accuracy and reliability of calibration; by collecting the pose information in different dimensions multiple times, the influence caused by errors can be reduced, improving the calibration accuracy.

[0064] After obtaining the coordinate information, step S200 can be executed.

[0065] S200: Obtain mechanical information according to each coordinate information, so as to calibrate the target product according to the mechanical information.

[0066] Further, the step S200 includes:

[0067] Input each pose image and the corresponding coordinate information into a preset neural network model to obtain mechanical information; use the actuator to position and fit the target product according to the mechanical information.

[0068] In this embodiment, 5 pose images and the corresponding coordinate information are input into a preset neural network model. Through the convolutional network, the pose images and the corresponding coordinate information are converted into mechanical information, so as to control the actuator to position and fit the target product according to the generated mechanical information.

[0069] In the above technical solution, by adopting a neural network model, the input pose images and coordinate information can be effectively processed and analyzed, so as to obtain accurate mechanical information, improving the efficiency and accuracy of calibration; the preset neural network model has automatic learning and adaptability, and can continuously improve the processing ability for various different coordinate information through training and optimization, adapting to different calibration scenarios; due to the fast calculation speed of the neural network model, real-time mechanical information acquisition and feedback can be realized, ensuring the application of the positioning and fitting of the target product in a fast production environment; adopting a neural network model can adapt to different target products. Whether the target product has shape changes, size changes, or pose changes, it can be accurately identified and positioned, improving the adaptability of the system; by automatically obtaining mechanical information and performing positioning and fitting by the actuator, the need for manual participation is reduced, and the complexity of the operation and the risk of human errors are reduced.

[0070] After completing the calibration of the target product, step S300 can be executed.

[0071] S300: Collect the image of the calibrated target product, and obtain the calibration result according to the target product image, so as to feedback compensation information according to the calibration result.

[0072] Furthermore, the collecting the image of the target product in step S300 includes:

[0073] Collect the positioned and fitted target product through the second vision module to obtain the target product image.

[0074] In this embodiment, for the positioned and fitted target product, another CCD vision system is used to collect image data to obtain the image of the target product, that is, the target product image.

[0075] In the above technical solution, by collecting the image of the positioned and fitted target product, the accuracy of the positioning and fitting can be verified to ensure that the target product is in the correct position; the collected target product image can be used for subsequent inspection and analysis. By comparing the differences with the ideal model, the positioning and fitting quality can be evaluated to further optimize the production process; and the image collection of the target product can be used for recording and archiving, forming a complete production process record, and serving as an important data basis for quality control and quality traceability.

[0076] Furthermore, the feedback compensation information in step S300 includes:

[0077] Compare the target product image with a preset image to obtain a deviation value; if the deviation value is greater than a preset threshold, it is determined that the current calibration of the target product is unqualified, and the target product image is fed back to the preset neural network model to perform data compensation according to the target product image; otherwise, it is determined that the current calibration of the target product is qualified, and the current process ends.

[0078] In this embodiment, the preset image is the pose image corresponding to the original position; compare the collected target product image with this preset image. If there is a deviation between the target product image and the preset image, the target product image is fed back to the preset neural network model to achieve automatic data compensation and form a closed-loop control.

[0079] Among them, the preset threshold is formulated through quality specifications, and specific values are not limited here.

[0080] In the above technical solution, by comparing with a preset image and calculating the deviation value, the system can automatically judge the calibration quality of the target product without manual intervention, improving the automation degree of the process; performing real-time deviation calculation and determination on the target product image can timely detect calibration problems, so as to take compensation measures in time and improve production efficiency; by judging whether the calibration is qualified according to the deviation value, the quality control of the production process can be carried out, and calibration anomalies can be detected and processed in time, improving the product quality level; by feeding back the target product image to the preset neural network model for data compensation, the calibration process can be continuously optimized, improving the accuracy and stability of calibration.

[0081] Embodiment 2:

[0082] Please refer to Figure 2 , this application also provides an automatic calibration and feedback system, and the system includes:

[0083] Acquisition module: used to acquire the coordinate information of the calibration block in a preset pose.

[0084] Further, the acquisition module includes:

[0085] First acquisition unit: used to acquire the preset pose images of the calibration block.

[0086] In this embodiment, under the first vision module, the calibration block is moved by the actuator and the preset pose is changed, so as to acquire the pose images corresponding to the preset pose in real time through the first vision module.

[0087] Among them, the preset number of poses is preferably 5, and those skilled in the art can also set it according to actual calibration requirements, not limited to only 5; the first vision module is a CCD vision system.

[0088] The actuator grabs the calibration block and moves it under the CCD vision system, and changes the pose 5 times through the actuator, including the rotation and movement of the x and y axes for operation, and the CCD vision system acquires the image information (i.e., the pose images) in different poses.

[0089] Second acquisition unit: used to obtain the pose information in the first dimension, the second dimension and the third dimension according to each pose image, so as to obtain the coordinate information corresponding to each pose image according to the pose information.

[0090] In this embodiment, the five pose images respectively include different pose information in the X, Y and R three dimensions to include all coordinate information.

[0091] The obtained coordinate information is, for example, P1(0, 0, 0), P2(-3, 3, 3), P3(3, 3, 3), P4(3, -3, -3), P5(-3, -3, -3); where P1 is the origin point.

[0092] In the above technical solution, by collecting the pose images corresponding to the coordinate information, the information of the calibration block in different poses can be accurately obtained, improving the accuracy of collection; the second collection unit can obtain the pose information in the first dimension, the second dimension and the third dimension, which can comprehensively reflect the position and pose information of the calibration block; through the automatic processing and analysis of the pose images, the automatic acquisition and calculation of the coordinate information can be realized, reducing the need for manual operation and improving the degree of automation.

[0093] Calibration module: used to obtain mechanical information according to the collection result of the collection module, so as to calibrate the target product according to the mechanical information.

[0094] Further, the calibration module includes:

[0095] The first acquisition unit: used to input each pose image and the corresponding coordinate information into a preset neural network model to obtain mechanical information.

[0096] Calibration unit: used to perform positioning and fitting on the target product according to the mechanical information.

[0097] In this embodiment, 5 pose images and the corresponding coordinate information are input into a preset neural network model, and the pose images and the corresponding coordinate information are converted into mechanical information through a convolutional network, so as to control the actuator to perform positioning and fitting on the target product according to the generated mechanical information.

[0098] In the above technical solution, through the preset neural network model and mechanical information, the calibration module can accurately calculate the position and pose information of the target product, and high-precision positioning and fitting can be realized; the preset neural network model has self-adaptability and learning ability, and can be automatically adjusted and optimized according to different calibration scenarios and data inputs, so as to adapt to various target products and calibration requirements; the calibration module uses the neural network model to automatically calculate and process the mechanical information, realizing the automatic operation of the calibration process, reducing the need for manual intervention, and improving the work efficiency; through the rapid calculation and adjustment of the calibration module, the time and cost of the calibration process can be reduced, and the efficiency of the production line can be improved.

[0099] Feedback module: used to collect the image of the calibrated target product, and obtain the calibration result according to the target product image, so as to feedback compensation information according to the calibration result.

[0100] Further, the feedback module includes:

[0101] The second acquisition unit: configured to acquire the target product after positioning and fitting to obtain an image of the target product.

[0102] In this embodiment, for the target product after positioning and fitting, another CCD vision system is used to collect image data to obtain an image of the target product, that is, the target product image.

[0103] The comparison unit: configured to compare the target product image with a preset image to obtain a deviation value.

[0104] The feedback unit: configured to, when the deviation value obtained by the comparison unit is greater than a preset threshold, feedback the target product image to the preset neural network model to perform data compensation based on the target product image.

[0105] In this embodiment, the preset image is the pose image corresponding to the original position; the acquired target product image is compared with this preset image. If there is a deviation between the target product image and the preset image, the target product image is fed back to the preset neural network model to achieve automatic data compensation and form a closed-loop control.

[0106] Wherein, the preset threshold is determined by quality specifications, and the specific value is not limited here.

[0107] In the above technical solution, the feedback module can quickly respond to the deviation of the target product, and improve the accuracy of positioning and fitting through data compensation operations; since the feedback module can collect the target product image in real time and perform comparison and judgment, the feedback processing has high real-time performance and fast response ability; the preset neural network model has self-adaptability and learning ability, and can automatically adjust and optimize according to the feedback data; the feedback module can quickly identify the deviation situation and perform data compensation processing, improving the efficiency and accuracy of the calibration system.

[0108] In summary, the present application provides an automatic calibration and feedback method and system. First, the coordinate information of the calibration block in a preset pose is collected. Then, mechanical information is obtained based on each coordinate information to calibrate the target product according to the mechanical information. Further, an image of the calibrated target product is collected, and a calibration result is obtained based on the target product image to feedback compensation information according to the calibration result. The method proposed in the present application has fewer calibration steps, a simple calibration process, and high calibration efficiency. At the same time, it can also reduce the calibration difficulty, improve the calibration accuracy, and form a closed-loop control with adaptive feedback, realizing real-time tracking and feedback of data and real-time adjustment of abnormal data. And the whole process is automatically completed, reducing manual intervention and operation errors, and greatly improving production efficiency and quality to a certain extent. In addition, this method is applicable to various different target products and production environments, and has high flexibility and versatility.

[0109] Although example embodiments have been described herein with reference to the accompanying drawings, it should be understood that the above example embodiments are merely exemplary and are not intended to limit the scope of the present application. Those of ordinary skill in the art can make various changes and modifications therein without departing from the scope and spirit of the present application. All such changes and modifications are intended to be included within the scope of the present application as claimed in the appended claims.

[0110] Those of ordinary skill in the art can realize that the units and algorithm steps of each example described in combination with the embodiments disclosed herein can be implemented by electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are executed in hardware or software depends on the specific application and design constraints of the technical solution. Professional technicians can use different methods to implement the described functions for each specific application, but such implementation should not be considered to exceed the scope of the present application.

[0111] In several embodiments provided by the present application, it should be understood that the disclosed devices and methods can be implemented in other ways. For example, the device embodiments described above are merely illustrative. For example, the division of the units is only a logical function division, and there may be other division methods in actual implementation. For example, multiple units or components can be combined or integrated into another device, or some features can be ignored or not executed.

[0112] It should be noted that in this text, relational terms such as first and second are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any actual relationship or order between these entities or operations. Moreover, the terms "comprising", "including" or any other variant thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device comprising a series of elements not only includes those elements, but also includes other elements not expressly listed, or further includes elements inherent to such process, method, article or device. Without further limitation, an element defined by the statement "comprising an..." does not exclude the presence of additional identical elements in the process, method, article or device comprising said element.

[0113] Although the description of the present application is made in conjunction with the above specific embodiments, it is obvious that those skilled in the art can make many substitutions, modifications and variations based on the above content. Therefore, all such substitutions, improvements and variations are included within the spirit and scope of the appended claims.

Claims

1. An automatic calibration and feedback method, characterized in that: The following steps are involved: S100: collecting coordinate information of the calibration block in a preset posture; wherein, before executing step S100, it includes: under the first visual module, moving the calibration block through the actuator and changing the preset secondary posture, so as to collect the posture image corresponding to the preset secondary posture in real time through the first visual module; The step S100 includes: acquiring posture information in the first dimension, the second dimension and the third dimension according to each posture image, so as to acquire coordinate information corresponding to each posture image according to the posture information; S200: Acquire mechanical information according to each coordinate information, so as to calibrate the target product according to the mechanical information; wherein, the step S200 includes: inputting each posture image and the corresponding coordinate information into a preset neural network model to acquire mechanical information; positioning and bonding the target product according to the mechanical information through the actuator; S300: collecting a calibrated target product image, and obtaining a calibration result according to the target product image, so as to feed back compensation information according to the calibration result; Among them, the step S300 includes: capturing the target product after positioning and bonding through the second visual module to obtain a target product image; comparing the target product image with a preset image to obtain a deviation value; if the deviation value is greater than a preset threshold, it is determined that the current target product calibration is unqualified, and the target product image is fed back to the preset neural network model to perform data compensation according to the target product image; otherwise, it is determined that the current target product calibration is qualified and the current process ends.

2. A system using the automatic calibration and feedback method as claimed in claim 1, characterized in that: The system comprises: Acquisition module: used to collect the coordinate information of the calibration block in a preset position; Calibration module: used for acquiring mechanical information according to the acquisition result of the acquisition module, so as to calibrate the target product according to the mechanical information; Feedback module: used to collect the calibrated target product image, and obtain the calibration result according to the target product image, so as to feed back compensation information according to the calibration result.

3. The system according to claim 2, characterized in that The acquisition module comprises: The first acquisition unit is used to acquire a preset posture image of the calibration block; The second acquisition unit is used to obtain the posture information in the first dimension, the second dimension and the third dimension according to each posture image, so as to obtain the coordinate information corresponding to each posture image according to the posture information.

4. The system according to claim 3, characterized in that The calibration module comprises: A first acquisition unit: used for inputting each posture image and corresponding coordinate information into a preset neural network model to acquire mechanical information; Calibration unit: used for positioning and fitting the target product according to the mechanical information.

5. The system according to claim 4, characterized in that The feedback module comprises: The second acquisition unit is used to collect the target product after positioning and bonding to obtain the target product image; Comparison unit: used for comparing the target product image with a preset image to obtain a deviation value; Feedback unit: used to feed back the target product image to the preset neural network model when the deviation value obtained by the comparison unit is greater than a preset threshold value, so as to perform data compensation according to the target product image.

Images